UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Visitor robert.bowen
Visitor
604 Views
Registered: ‎12-06-2017

Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

Hello,

I am working on a design with the Xilinx Radiation Text Consortium (XRTC).  The AC Coupling caps were dropped during the routing of the board by mistake.  We are being told that adding them into the design will require a complete re-do of all board routing.  The clocks do have the AC coupling capacitors, but the RX/TX data lanes do not.  What are the implications if the board is built without the AC Coupling caps?  We do want to test the MGT lanes.  Are there any workarounds?  The board has currently been routed and has the associated gerber and ODB++ files.

 

Thanks,

--Bobby

0 Kudos
1 Solution

Accepted Solutions
502 Views
Registered: ‎09-17-2018

Re: Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

I know,

I reviewed the layout.  No problem.  Say hi to Gary for me.  No caps needed.

l.e.o.

View solution in original post

0 Kudos
5 Replies
559 Views
Registered: ‎01-08-2012

Re: Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

Three issues:

Rail Sequencing.  You did not say whether the driver and receiver have the same power supply.  (You didn't even say whether they were on the same board, but I'm going to assume they are.)  If they have different power supplies, then it's possible that the driver can be powered up while the receiver is powered down.  This is unlikely to destroy anything (due to the current limiting effect of the CML outputs' 50 ohm pullup resistances), but it's a situation you should try to avoid.

Protocol issues.  Some protocols designed for hot swap (PCIe, SATA, etc.) do funny things to detect whether there's something plugged in at the other end.  You didn't mention the protocol, but if it needs the caps to work, you might be out of luck.

Common mode voltage.  If the output common mode voltage of the driver is within the acceptable range for the receiver, then it's probably going to work.  The Xilinx receiver's self bias is typically turned on (because almost everyone uses AC coupling), but you could consider either turning it off (using RX_CM_SEL) or setting it to a voltage that's compatible with the driver (using RX_CM_TRIM).  If the common mode voltages (as specified in the datasheets) aren't compatible, you might get lucky and find that it works anyway (as sometimes the datasheet figures are rather conservative).  If it doesn't work, you have plenty of time to rethink the design processes that let you get into this situation while you wait for the next batch of boards from the manufacturer.

More outlandish things you can do to fix a common mode voltage issue might include trimming the MGTAVTT voltage on the board (possibly taking it outside the recommended range) or adjusting the TXDIFFCTRL setting on the transmitter output.

 

EDIT:

Look at UG476 Figure 3-29 (Tx) and Figure 4-2 (Rx).

Read DS183 Table 53.

Moderator
Moderator
533 Views
Registered: ‎07-30-2007

Re: Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

As Allan says the common mode voltage of the transmitter is going to the important.  There is an app note on DC coupling the 7 Series GTX if that is the transceiver you are using: https://www.xilinx.com/support/documentation/application_notes/xapp1096-dc-coupling-gtx-transceivers.pdf




----------------------------------------------------------------------------
Don't forget to reply, kudo, and accept as solution
----------------------------------------------------------------------------


0 Kudos
528 Views
Registered: ‎09-17-2018

Re: Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

r.b,

Do the MGT's driving the receive side reside on the same part?  Do they come from a different board, different device?

AC coupling is used when the drivers are other than Xilinx devices and whose common mode voltages (Vcm) are not compatible with the Xilinx device receiving.

If the drivers have a Vcm that is within specification, then no caps are needed.

l.e.o.

0 Kudos
Visitor robert.bowen
Visitor
507 Views
Registered: ‎12-06-2017

Re: Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

Thanks for the quick responses.  The app note was extremely helpful and it looks like there is saving grace after all.

More information on this board:  The board has been designed to radiation test different primitives on the Virtex-7.  In the setup, there are two V7s with very similar loads talking to one another on separate DUT cards connected by a backplane (using the same power supplies).  One of the DUT cards get exposed to radiation and the other is outside of the beam.  The motherboard routes corresponding IO from one DUT card to the other (LVCMOS to LVCMOS, LVDS to LVDS, MGT to MGT, etc).  Therefore the TX lanes of a given bank are routed to the RX of the same bank number on the other V7 (matching V7 PNs).  The protocols being used are Aurora and RadioIO.

0 Kudos
503 Views
Registered: ‎09-17-2018

Re: Virtex-7 Effects of MGTs without AC Coupling

Jump to solution

I know,

I reviewed the layout.  No problem.  Say hi to Gary for me.  No caps needed.

l.e.o.

View solution in original post

0 Kudos